The Postdoctoral Research Scientist will serve as a leading technical contributor to the federally-funded Mosaic Inference Defender against LLM-Assisted Threats (MIDLLMAT) project, driving the design, development, and evaluation of an AI system to protect sensitive government information. The position will focus on one of two core system components, the Filter or the Sensor, taking responsibility for its technical
development and conducting the testing required to demonstrate that performance meets the funder's standards. The incumbent will work under the direction of the senior Principal Investigators as part of a team that includes postdoctoral researchers, graduate and undergraduate students, and external consultants. Key responsibilities will include leading design and evaluation activities for the assigned component,
applying current methodological approaches to system assessment and benchmarking. The role encompasses developing and executing evaluation protocols, analyzing results against the performance criteria specified by the funder, preparing technical reports and peer-reviewed publications that document the rigor of the work, and finalizing software and data artifacts and documentation for transfer to the funder. The
incumbent will exercise substantial independence in planning, executing, and communicating research results, while coordinating closely with other members of the team on timelines, deliverables, and system integration, as well as contributing to additional aspects of the project as needed. This position is for the first phase of the project and may be extended by an additional 6 months at the funder’s discretion.
Summary from the source posting. Always confirm details on the institution's official career page.